1. Field of the Invention
The present invention relates to a communication network system for executing data exchange among a plurality of nodes connected onto a transmission line.
2. Description of the Prior Art
In trials for higher-speed local area networks (hereinafter, referred to as LAN) that have been put into practical use so far, parallel access of a LAN with two or more nodes, i.e. spacial reuse of the transmission line is not permitted from the viewpoint of architectural simplification, as is typified by FDDI(fiber distributed data interface). However, adopting such a method would involve the following problems. Firstly, only a single node is allowed to access the network at all times. As a result, the throughput of the system is smaller than that for the bandwidth of the transmission line. Secondly, wider bandwidths lead to lower efficiency. Thirdly, greater size of LANs would take a longer time to acquire a right of access to the transmission line, which would result in increased delay of data transmission. To solve these problems, various types of communication network systems have been proposed in recent years.
Japanese Patent Laid-Open Publication SHO 63-240148 has proposed a communication network in which a plurality of nodes are connected to one another by a transmission line in the form of ring, where transmission frames divided into a plurality of transmission areas are circulated on the transmission line. Tokens are assigned to the transmission areas, respectively. For data transmission, each node acquires any one of the tokens and outputs data to its corresponding transmission area. Thus, spacial reuse of the transmission line is implemented.
In this patent, however, the length of data transmitted at a time is limited by the length of the transmission area. Also, when only a small number of nodes less than the number of tokens are transmitting data, some transmission areas are out of use, inefficiently. In particular, when the number of nodes on the transmission line is less than the number of tokens, part of the transmission line is at all times out of use. Reducing the number of tokens to avoid this would result in a lower throughput. Besides, this patent could not ensure the fairness for the nodes to access the transmission line.
Japanese Patent Laid-Open Publication HEI 3-205936 (U.S. Pat. No. 4,926,418) has proposed a system in which a dual ring is assumed as a transmission line, where the nodes utilize a buffer insertion access control mechanism. Each node on the transmission line has a frame insertion buffer and is able to transmit a packet onto the transmission line as far as the frame insertion buffer is empty. Any packet that has been transmitted from an upperstream node during data transmission and destined to any other than the node is temporarily stored in the frame insertion buffer, and relayed onto the transmission line at the time point when my-node packet transmission has been completed. Further, the quantity of packets that can be transmitted by each node at a time is limited by using a control message circulating in a direction opposite to the direction of packet transmission so that the fairness over the whole network can be maintained. The maximum number of frames (hereinafter represented by k) that can be transmitted by the time when one node observes two successive control messages as well as the number of frames (hereinafter represented by 1) that can at least be transmitted between two successive control messages are determined.
This patent, however, takes the dual ring as a presumption, so that it cannot be utilized with a single ring. Moreover, when a control message is lost, more than k number of packets cannot be transmitted by a time-out. Further, once a control message is lost, a large number of nodes generate control messages, with the result that a plurality of control messages flow on the transmission line. When a node holding a control message receives another control message, the newly coming control message is discarded so that the control messages are settled finally into one. However, since the node reacts with control messages that come up one after another before the settlement, the packet transmission is blocked to one in number. Furthermore, this system uses a method different from the conventional access control method using tokens, thus having difficulty in succeeding the existing techniques.